Gartner predicts ‘digital twins of a customer’ will transform CX

Were you unable to attend Transform 2022? Check out all of the summit sessions in our on-demand library now! Watch here.

Digital twins of physical products and infrastructure are already transforming how companies design and manufacture products, equipment and infrastructure. In its latest Immersive Hype Cycle, Gartner predicts that digital twins of a customer (DToC) could transform the way enterprises deliver experiences. Simulating a customer experience (CX) is a bit more nuanced than a machine — and there are privacy considerations to address, not to mention the creepiness factor. Though if done right, Gartner predicts the DToC will drive sales while delighting customers in surprising ways. 

Gartner has a nuanced view of the customer, including individuals, personas, groups of people and even machines. It is worth noting that many enterprise technologies are moving toward this more comprehensive vision. Customer data platforms consolidate a data trail of all aspects of customer interaction. Voice of the customer tools help capture data from surveys, sensors and social media. While, customer journey mapping and customer 360 tools analyze how customers interact with brands across multiple apps and channels. 

The critical innovation point of DToC is that it helps contextualize data to help understand what customers really need to improve the overall experience, Gartner VP analyst Michelle DeClue-Duerst told VentureBeat. For example, a hotel with knowledge about a customer’s gluten allergy might identify nearby gluten-free restaurants and only stock the minibar with snacks the customer will enjoy. 

When done right, DToCs can help business teams design ways to serve or capture customers and facilitate new data-driven business models. They will also improve customer engagement, retention and lifetime failure. 


MetaBeat 2022

MetaBeat will bring together thought leaders to give guidance on how metaverse technology will transform the way all industries communicate and do business on October 4 in San Francisco, CA.

Register Here

Developing core capabilities

Gartner notes that DToC implementations are still embryonic, with about 1-5%penetration of the target audience. At the same time, enterprises have been busy finding ways to get the most value from their investment using various marketing analytics tools. 

Subha Tatavarti, CTO, Wipro, told VentureBeat there have been several important milestones in using tools for simulating customers to improve experiences. The most notable have been the ability to define customer experience transformation objectives, including the capability to identify and assess data assets, personas and processes and tools for building and testing behavior models. New ModelOps approaches for integrating monitoring and enhancing the models are also advancing the field.

“A new generation of recommendation systems based on intention, context and anticipated needs is a very exciting development in combined modeling and simulation capabilities,” Tatavarti said. “Personalized learning and hyper-personalized products are great advancements and personalized healthcare will have critical impacts on that industry.”

Enterprises are taking advantage of new identity resolution capabilities that assemble pieces of data to create a holistic view of the customer. This stitching can help a company understand what an individual customer buys, how frequently they purchase, how much they spend, how often they visit a website and more. 

“Without identity resolution, the company may have to rely on only some of the attributed data sources to fill out the digital persona, meaning the simulation would be somewhat inaccurate,” said Marc Mathies, senior vice president of platform evolution at Vericast, a marketing solutions company.

Bumpy road

Enterprises will need to address a few challenges to scale these efforts. Gartner observed that privacy and security concerns could lengthen the time it takes DToCs to mature and increase regulatory risks. Organizations must also build teams familiar with machine learning and simulation techniques. 

Tatavarti said the most difficult obstacles are the quality and availability of customer data from physical and digital interaction and data sharing between multiple organizations. These challenges will also involve privacy considerations and the ability to connect physical systems and virtual models without affecting the experience or performance. Teams also need to ensure the accuracy of the models and eliminate bias.

Bill Waid, chief product and technology officer at FICO, a customer analytics leader, told VentureBeat that another challenge in implementing digital twins for customer simulation is the impact of localized versus global simulation. Frequently, teams only simulate subsegments of the decision process to improve scale and manageability. Enterprises will need to compose these digital twins for more holistic and reusable simulations.

Organizations will also need to be transparent. 

“Initially, it will be hard to convince customers they need a digital twin that your brand stores and that the customer should help create it to improve their experience,” said Jonathan Moran, head of MarTech solutions marketing at SAS.

Building the right foundation

Industry leaders have many ideas about how enterprises can improve these efforts. 

Unlike digital twins in areas like manufacturing, customer behavior shifts quickly and often, Karl Haller, partner at IBM Consulting said it is essential to implement ongoing optimization and calibration to analyze the simulation results and determine ways to improve the performance of the models. He also recommends narrowly defining the focus of a customer simulation to optimize outcomes and reduce costs. Innovations in natural language processing, machine learning, object andvisual recognition, acoustic analytics and signal processing could help. 

Moran recommends enterprises develop synthetic data generation expertise to build and augment virtual customer profiles. These efforts could help expand data analytics and address privacy considerations.

Mark Smith, vice president of digital engagement solutions at CSG, recommends business to overlay voice of customer data with behavioral data captured through customer journey analytics. This modeling method is typically the fastest and most accurate route to understanding the peaks and valleys of the customer journey. 

“Comparing customers’ actual actions with their reported lived experience data unearths disconnects between customers’ perception of the experience and brands’ analysis of their own offerings,” Smith said. 

A mixed future 

Eventually, enterprises will need to find ways to optimize for profits along with customer well-being. Eangelica Germano Aton, product owner at a conversational intelligence platform, Gryphon AI, predicts that things will initially get worse for people as machines get better at predicting choices that reduce emotional well-being. 

“I think it will take a customer-driven or a bottom-up revolution and rejection of the current model before a more sophisticated and genuinely humanist AI can emerge that doesn’t maximize such a shallow objective function as profit,” Germano Aton said. 

Others are more optimistic. 

“Over time, it will be possible to use a deep understanding of the customer in a way that creates value for the consumer, the brand and the employees of the brand,” said Chris Jones, chief product officer at Amperity, a CDP platform. “One of the things we are observing is the ability of these capabilities to deepen the human connection between brands and the customers they serve by empowering employees across the brand to truly see their customer and provide the most personalized experience possible.”

In the long run, digital twin capabilities could become embedded into marketing and customer experience automation tools.

“As digital twin work moves more into marketing and CX in five to ten years, I think we will see solutions with more simulation capabilities built in,” Moran said. “Any type of marketing KPI and expected results will be simulated within the tool. Vendors already have some simulation capabilities for optimization, reinforcement learning and predictions, but I think this will start to increase even more in the coming years.”

Originally appeared on: TheSpuzz